AVS1996 Session TF-TuM: Modeling of Thin Film Deposition

Tuesday, October 15, 1996 8:20 AM in Room 107B

Tuesday Morning

Time Period TuM Sessions | Abstract Timeline | Topic TF Sessions | Time Periods | Topics | AVS1996 Schedule

Start Invited? Item
8:20 AM TF-TuM-1 Monte Carlo Simulations of Nucleation in Rectangular Trenches
A. Challa, D. Wang, T. Cale (Arizona State University)
The density of islands nucleated during deposition inside ideal rectangular trenches is investigated using Monte Carlo simulations. The depositing species is assumed to have a sticking factor of one. Simulated island density and average island size both decrease along the wall from top to bottom and increase from the well-base corner to the center along the base, reflecting the spatial dependence of the flux in the trench. As the trench aspect ratio increases, the nucleation density and average island size at the corner and center along base become closer together. Implications for metallization applications are discussed.
9:00 AM TF-TuM-3 Monte Carlo Simulation of Thin Film Deposition by Supersonic Molecular Beams
G. Chen, I. Boyd (Cornell University)
The use of supersonic molecular beams as sources for thin film growth represents a novel approach to film deposition. The seeding technique employed allows considerable control over the incident kinetic energy of the reactant molecules striking the substrate. In particular, hyperthermal kinetic energy greater than 1eV can be readily achieved. It has been demonstrated experimentally that the surface reaction probability can be greatly enhanced by increasing kinetic energy. The current study considers expansion of a mixture of 1% Si\sub 2\H\sub 6\ and 99% H\sub 2\ from a nozzle orifice, through a conical skimmer, and into the growth chamber. Silicon thin film is deposited over the substrate surface. Due to the rapid decrease in density, the gas varies from continuum flow to near free molecular flow. The direct simulation Monte Carlo method (DSMC) is employed to simulate this flow with large variation of length scale. This particle based method provides detailed flow field information in the deposition chamber and is expected to aid in optimization of the system design. The nozzle temperature dependence of disilane impact kinetic energy over the substrate surface is studied. A strong dependence will imply easy control over thin film growth rate by varying the nozzle temperature. The peak incident energy is compared with experimental measurement. Results are presented for disilane incident energy, impact angle and flux distributions across the substrate to study the uniformity of the thin film. Reactant velocity distribution is also compared with time-of-flight measurements.This work is supported by the National Science Foundation.
9:20 AM TF-TuM-4 3 Layer Formalism for the Description of Ge Segregation during SiGe Alloy Growth by Molecular Beam Epitaxy
D. Godbey, M. Ancona (Naval Research Laboratory)
The growth and capping of SiGe alloy layers by solid source molecular beam epitaxy was recently reported and modeled using a static 2-layer description by this group\super 1\. The static 2-layer formalism lends itself to an analytical solution of the differential equation; however, the assumptions that growth can be treated 2 layers at a time and that the segregation rate is much faster than the growth rate, fails to adequately describe the dependence of x for growth of Si\sub 1-x\Ge\sub x\ and alloy growth at low temperature. In this work, the modeling of Ge segregation during Si\sub 1-x\Ge\sub x\ growth was expanded to allow for simultaneous growth and segregation. Film growth of a single layer with concurrent segregation requires the following of 3 layers simultaneously (3-layer model). After partial growth of the topmost layer, the surface region consists of the top layer and the uncovered fraction of the second layer. Thus, surface segregation is observed between the topmost and second layers, and the uncovered fraction of the second layer with the third layer. Exchanges between the second and third layers continue until the second layer is fully covered. The best fit with experimental data for the 3-layer model was obtained with fitting parameters of 1.7 and 0.14 eV for the activation barrier and segregation energies respectively. This is in comparison to the modeling parameters obtained for the 2-layer model of 1.63 and 0.28 eV respectively\super 1\. \super 1\ Godbey and Ancona, J. Vac. Sci. Technol. B 11, 1120 (1993)
9:40 AM TF-TuM-5 A Monte Carlo Simulation of the Physical Vapor Deposition of Nickel
Y. Yang, R. Johnson, H. Wadley (University of Virginia)
A two-step Monte Carlo method for atomistically simulating physical vapor deposition processes is developed and used to model the two-dimensional physical vapor deposition of nickel. The method consists of an impact approximation for the initial adatom adsorption on a surface and a multipath diffusion analysis to simulate subsequent surface morphology and interior atomic structure evolution. An embedded atom method is used to determine the activation energies for each of the many available diffusional paths. The method has been used to predict the morphology/structure evolution of nickel films over the length and time scales encountered in practical deposition processes. For planar substrates, the modeling approach has enabled determination of the effect of vapor processing variables such as flux orientation, deposition rate, and substrate temperature on deposit morphology/microstructure as defined by relative density, surface roughness and growth column width (which appears closely related to grain size). It has also been used to simulate a trench-filling metallization process under different conditions.
10:00 AM TF-TuM-6 Microstructural Studies and Simulation of Films Sputtered onto Non-planar Substrates
M. Brett (University of Alberta, Canada)
When thin films are sputtered over non-planar topography on a substrate, the microstructural properties of the films (ie: porosity, grain size, orientation, composition) may vary greatly within the topographical feature. The major factors creating this change are the incident species' angular and energy distributions at the substrate and, particularly for high temperature deposition, minimization of surface and interface energies. This talk will present both experimental and simulation studies of film microstructure and film coverage in topography for films deposited using many variations of sputtering, including bias sputtering, ionized sputtering, long throw- low pressure sputtering, collimated sputtering, and sputtering onto elevated temperature substrates. Features studied include sidewall and bottom film coverage and quality within topography, the "breadloafing" effect at the opening of vias, and process variability in high temperature or reflowed films created by the random nature of grain growth. Simulations will be presented using the microstructure and film growth model SIMBAD, and the new kinetic-based and grain growth simulator GROFILMS. It will be demonstrated that although current deposition simulators are very powerful, more experimental studies of the details of atom-surface interactions are necessary to fully understand current sputter technologies and associated film properties.
10:40 AM TF-TuM-8 Across-Wafer Nonuniformity of Long-Throw Sputter Deposition
A. Mayo, S. Hamaguchi (IBM T.J. Watson Research Center); J. Joo (Kunsan University, Korea); S. Rossnagel (IBM T.J. Watson Research Center)
Due to the finite size of the sputtering target, the angular distribution of sputtered neutral particles (atoms) generally varies on the wafer surface from its center to edge. This effect is somewhat mitigated in long-throw sputtering deposition systems, where the directionality of the incoming neutral particles is high due to the narrow view angle toward the target from the substrate position. However, even in long-throw deposition systems, if the sizes of the target and wafer are comparable, the left-right asymmetry of deposited film thicknesses over trench side walls located near the edge of the wafer may be significant. We use numerical simulation as well as metal sputter deposition experiments to study this effect for trenches with various aspect ratios located at various positions in the system. The numerical simulations are performed using SHADE, the micro-profile evolution code based on the shock-tracking algorithm. We shall also discuss how ionized magnetron sputter deposition systems can circumvent this non-uniformity problem.
11:00 AM TF-TuM-9 Enhanced Diffusion Rates of Two-dimensional Pt Clusters on Pt(111)
V. Chirita, E. Munger (Link\um o\ping University, Sweden); J. Greene (University of Illinois, Urbana); J. Sundgren (Link\um o\ping University, Sweden)
Molecular dynamics were used to investigate the dynamics of the motion of hexagonal Pt\sub 7\ and Pt\sub 6\ clusters on Pt(111). It was found that close-packed heptamers on fcc terrace sites are very stable structures, exhibiting only rarely reconfiguration or translation events, over simulations times > 40 ns at 1000K. The Pt\sub 6\ atomic structure investigated was that rendered by the removal of central atoms from the regular, close-packed Pt\sub 7\ clusters. The presence of the central vacancy in clusters induced significant increases in surface mobilities of depleted heptamers. Specifically, shape changes, rotations and/or translations over several lattice sites of cluster's center of mass (CM) are observed for time scales < 1 ns, at the same temperature of 1000 K. Typical mechanisms accountable for cluster migration are presented and discussed. The CM average velocity of depleted clusters is compared to that of close-packed structures and found to increase by more than an order of magnitude. These results expand on our very recent work\super 1\, showing that the adsorption of single adatoms on close-packed Pt\sub 7\ clusters, also enhances cluster diffusion rates. The relevance of these findings to thin film growth kinetics will be discussed. \super 1\ E.P. Munger, V. Chirita, J.E. Greene and J.-E. Sundgren, Surf. Sci. 355, L325 (1996).
11:20 AM TF-TuM-10 Surface Smoothing with Energetic Cluster Beams
Z. Insepov (Kazakh Polytechnical Institute, Kazakhstan); I. Yamada (Kyoto University, Japan)
Surface smoothing is one of the remarkable effects of impacts of energetic clusters, consisting of hundreds of atoms. To understand its mechanism we have developed a hybrid model, which utilizes Molecular Dynamics (MD) to simulate rapid colliosional processes at the central impact zone, and a finite-difference method to account for thermal evaporative processes occurring on a longer time scale over wider target area. A case of argon cluster of a hundred atoms impacting a silicon target with the energy of a few keV has been considered. MD simulation utilizes Buckingham two body potential to represent Ar-Ar and Ar-Si interactions and Stillionger-Weber potential for Si-Si interactions. Surface profile is described by the noisy Kuramoto-Sivashinsky equation and a Monte Carlo (MC) procedure is used for handling crater formation.MD simulation revealed that atoms ejected from the surface have a significant lateral momentum component (parallel to the surface) and that this has a major effect on surface smoothing when the initial cluster momentum is perpendicular to the surface (normal impacts). The angular distribution of atoms ejected after the impact, however, is modified by the presence of the evaporative component, which follows the usual cosine law. The impact zone remains hot enough to allow for a significant contribution of the latter effect. In contrast to the normal impacts, oblique impacts of clusters results in strongly asymmetrical distributions of atoms ejected from the surface, formation of asymmetric craters, and increased surface roughness. Comparison of the results of the simulations with experimental data shows qualitative agreement.
11:40 AM TF-TuM-11 Thin Film Growth via Low Energy Cluster Deposition: A Molecular Dynamics Study
C. Kelchner (Iowa State University\super 1\); A. DePristo (Iowa State University)
There is a great deal of interest in depositing clusters of atoms instead of single atoms to grow thin films. Two different results are plausible for cluster deposition. One is that the clusters create a good, strongly adhering thin film, as in high energy cluster deposition where the energy is localized on the surface up on cluster adsorption. Another possibility, particularly for low energy cluster deposition, is that the clusters retain some of their original structure and properties, leading to new types of materials. Experimental evidence exists for both situations.Molecular dynamics (MD) simulations permit multiple-layer thin film growth to be studied in detail, using reliable interatomic potentials for fcc metals from corrected effective medium theory. We will present MD simulation results for the low energy deposition of 5- and 10-atom clusters during growth of 20 layers of Pd on Pd(001) and Cu on Cu(001) at 80 K. The growth of these cluster-deposited thin films is much rougher than that of films grown via single atom deposition. The increased surface roughness can be attributed to two factors: (1) most deposition events add atoms to two or more layers; and (2) the growth of (111) facets on the surface produces many partially exposed atoms. Neither of these features was observed during the deposition of single atoms. In accord with these two factors, thin films grown by deposition of larger clusters tend to be rougher than those produced by smaller clusters. Furthermore, the higher coordination of the atoms in larger clusters means that they are more likely to retain the original cluster structure upon adsorption.This work was supported by NSF grant CHE-9224884.\super 1\Present address: Sandia National Laboratory, Livermore, CA
Time Period TuM Sessions | Abstract Timeline | Topic TF Sessions | Time Periods | Topics | AVS1996 Schedule